- Bad fraud practices are pervasive
- Attention to a short list of bad practices can bring improvement
- Banks must stay consistent and constant
The always-informative Frank on Fraud blog discusses -- with accompanying highly entertaining imagery and GIFs -- the 10 Worst Fraud Practices and How to Avoid Them. Frank recognizes that the same mistakes seem to be made "year in and year out," so he put together a handy list to aid in addressing and tackling them.
It’s like a rollercoaster. Make a mistake. Fix the problem. Make the same mistake again. Fix the same problem.
Top 10 Fraud Worst Practices
McKenna notes that millions and millions of dollars in losses can be avoided each day due by avoiding bad practices within bank's fraud prevention programs.
- Turning Off Fraud Strategies
- Buying A Fraud Tool and Letting It Run Itself
- Creating Too Many Rules and Overloading Your Operations
- Getting Rid of Fraud Staff When Losses are Low
- Hiring a Bad Fraud Manager
- Ignoring Fraud That is Written Off as a Credit Risk
- Allowing a Toxic Sales Culture to Run Amuck
- Thinking Everything is A-Ok
- Keeping an Old Fraud Technology Just Because You Understand it
- Putting Fraud in the Wrong Place Organizationally
Worst Practices for Check Fraud
When we take a closer look at the list, we can identify practices that are particularly harmful and in need of addressing for check fraud.
1. Turning Off Fraud Strategies
Believe it or not, this happens all the time. Executives of a bank will receive escalated calls from customers complaining about how they were impacted by the fraud strategies – like their card was declined.
The executives will demand that the fraud strategies be reduced to avoid impacting good customers. The fraud managers comply by turning off high volume fraud strategies.
The only problem is that those fraud strategies also PREVENTED LOTS OF FRAUD. So when they are turned off – fraud losses spike.
This is a problem with check fraud, as the fraud detection tools will flag check items for manual review -- holding up payments until a fraud analyst is able to deny payment or clear the check item. With image-forensics AI, the feedback provided by fraud analysts making the decision on a good or bad item helps build a stronger profile -- speeding up the process. Turning this off weakens the system and enables more fraud to get through.
2. Buying A Fraud Tool and Letting It Run Itself
Fraud tools are just that – tools. They can help you identify fraud but they can’t run themselves. Sadly, one mistake banks make, is writing a big check for a Fraud Solution, and then expect that it will run itself..
This mistake builds off of the first mistake. Fraud tools -- particular for check fraud -- need feedback. As we noted, each time a check item is cleared or denied, the AI system builds a stronger profile. This will lead to increased detection capabilities while reducing the number of false positives.
3. Creating Too Many Rules And Overloading Your Operations
Trust your fraud scores and analytics. You should rely on models 80% of the time. Rely on expert rules 20% of the time. And never, never have so many rules that you don’t know what they are all doing. If you have more than 100 rules in the system, it might be too much.
A mistake that many banks make is creating too many rules or maintaining unrealistic thresholds within check fraud systems. As banks tune their check fraud systems, it's important to understand how much fraud their operations team can realistically review, then adjust the system accordingly.
9. Keeping an Old Fraud Technology Just Because You Understand it
I see some really old, old technology at banks that are used to fight fraud. Check fraud is a good example of that. Many banks are still using technology built in 1995 (right around the time that Windows 95 launched) to detect their check fraud losses.
Why do they still use it? Because they understand it. And they are afraid of new technology and don’t want to make the leap.
How to Avoid It
Don’t be a prisoner of your old technology. Take a leap of faith with models, machine learning and all of the progress that has been made over the last 20 years. Get off those old Windows 95 machines and into the much newer and much better technology available.
Many banks are planning and currently deploying AI systems like image-forensic AI to replace these outdated systems. This, combined with new transaction-based AI systems, are providing a higher level of check fraud detection that can effectively reduce fraud losses.
Consistent and Constant
The key message that Frank on Fraud delivers is consistency. Many financial institutions put fraud prevention measures in place, only to eventually rely on a sort of "auto-pilot" to keep them on course. It's necessary to maintain a hands-on approach, staying up-to-date and consistent across the board in order to maintain success.
It's also recommended that banks deploy the latest technologies, such as transactional and image-based systems leveraging AI, to fight fraud.
Leveraging AI and machine learning for transaction-based systems enables banks to analyze millions of transactions, spotting anomalous transactions and behaviors that indicate fraudulent items. This is complemented by image forensic AI, which analyzes and interrogates the check images to detect counterfeit, altered, and forged checks.
Together, these systems -- and avoiding the worst practices -- protects banks and their customers from unnecessary losses.